Organizers

• Mireille Hildebrandt (Vrije Universiteit Brussels)
• Dr. Gianmarco Gori (Durham University)

Event Contact

• Mireille Hildebrandt (Vrije Universiteit Brussels)

Background

Organizers

• Mireille Hildebrandt (Vrije Universiteit Brussel)

Event Contact

• Gianmarco Gori (Vrije Universiteit Brussel)

Programme

Background

In this tutorial we will focus on the extent to which Generative AI, based on ‘Large Whatever Models’, falls within the scope of the AI Act and on the kind of legal obligations that should be taken into account by the developers of Generative AI that is meant to contribute to human-centric AI.

To this end we will first unpack the legal definitions of General Purpose AI Models (GPAI Models) and General Purpose AI Systems (GPAI Systems) and explain what kind of models qualify as GPAI models and what kind of systems qualify as GPAI systems. This will be followed by an inquiry into when a GPAI system is – legally speaking – a high risk AI system and into when a GPAI model is – legally speaking – an AI model generating systemic risk.

Second, we will elicit a small set of requirements that must be met by providers and/or deployers of GPAI Systems that integrate GPAI Models. As the HAI-NET is focused on contributing to real world human-centric AI, we will not focus on the research exemption that may apply to HAI-NET research. The whole point of legal protection by design is to ensure that such protection is built into the design phase. This means that developers must be aware of the requirements that providers and/or deployers of real-world applications of their models face.

Finally, we need to emphasise that our objective is to give our audience a first taste of the legal regime that applies to real world human-centric AI systems that integrate generative AI. For more an in-depth understanding we refer to the HAI-NET report that Dr. Gori is preparing on the subject and to the Chapter that Dr. Gori and Prof. Hildebrandt are co-authoring in the Handbook of Generative AI for Human-AI Collaboration, eds. Mohamed Chetouani, Andrzej Nowak and Paul Lukowicz (Springer forthcoming).

This project aims to take seriously the fact that the development and deployment of AI systems is not above the law, as decided in constitutional democracies. This feeds into the task of addressing the question of incorporation of fundamental rights protection into the architecture of AI systems including (1) checks and balances of the Rule of Law and (2) requirements imposed by positive law that elaborates fundamental rights protection.

A key result of this task will be a report on a coherent set of design principles firmly grounded in relevant positive law, with a clear emphasis on European law (both EU and Council of Europe). To help developers understand the core tenets of the EU legal framework, we have developed two tutorials, one in 2020 on Legal Protection by Design in relation to EU data protection law [hyperlink to Tutorial 2020] and one in 2021 on the European Commission’s proposal of an EU AI Act [hyperlink to Tutorial 2021]. In the Fall of 2022 we will follow up with a Tutorial on the proposed EU AI Liability Directive.

Our findings will entail: - A sufficiently detailed overview of legally relevant roles, such as end-users, targeted persons, software developers, hardware manufacturers, those who put AI applications on the market, platforms that integrate service provision both vertical and horizontal, providers of infrastructure (telecom providers, cloud providers, providers of cyber-physical infrastructure, smart grid providers, etc.);

A sufficiently detailed legal vocabulary, explained at the level of AI applications, such as legal subjects, legal objects, legal rights and obligations, private law liability, fundamental rights protection; - High level principles that anchor the Rule of Law: transparency (e.g. explainability, preregistration of research design), accountability (e.g. clear attribution of tort liability, fines by relevant supervisors, criminal law liability), contestability (e.g. the repertoire of legal remedies, adversarial structure of legal procedure).

Organizers

• Roberto Trasarti (CNR)
• Valerio Grossi (CNR)

Event Contact

• Chiara Boldrini (CNR)

Programme

Objectives

The objectives of the tutorial are to show how SoBigData RI can support data scientists in doing cutting-edge science and experiments. In this perspective, our target audience also includes people interested in big data analytics, computational social science, digital humanities, city planners, wellbeing, migration, sport, health within the legal/ethical framework for responsible data science and artificial intelligence applications. With its tools and services, SoBigData RI promotes the possibilities that new generations of researchers have for executing large-scale experiments on the cloud making them accessible and transparent to a community. Moreover, specialized libraries developed in SoBigData++ project will be freely accessible in order to make cutting edge science in a cross-field environment.

Format: The tutorial will be 3 hours containing:

• 1 hour of presentations describing the European project SoBigData++, the RI Services, and the Responsible Data Science principles and tools;
• 45 minutes and half of practical use of the RI with real examples of analysis in a dedicated Virtual research environment;
• 20 minutes for an open discussion with the attendees on the various aspects presented.